[ 
https://issues.apache.org/jira/browse/HIVE-24717?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17277535#comment-17277535
 ] 

Mustafa İman commented on HIVE-24717:
-------------------------------------

This caused so many precommit tests to fail. Why is below
 # Majority of tests run on local file system which is a ChecksumFileSystem.
 # ChecksumFileSystem#listStatusIterator() does not ignore checksum files 
(.A.crc files)
 # listStatusIterator's next returns A first, .A.crc second.
 # .A.crc files are automatically deleted by ChecksumFileSystem when file A is 
deleted.
 # Since we delete file A before iterator processes .A.crc, iterator cannot 
find .A.crc when it checks its permissions.
 # On Linux and Macos, LocalFileSystem invokes "ls -ld .A.crc" to get the 
permissions.
 # This exits with code 2 on Linux and iterator just throws.

Interestingly, the tests run fine on my computer. The reason is I use a Mac. 
"ls -ld" returns 1 for the same error on Macos. Hadoop ignores exit code 1 but 
not 2. So the tests pass on Mac but fail on Linux.

Apparently this is fixed on Hadoop 3.2 via 
https://issues.apache.org/jira/browse/HADOOP-12502 . We are using Hadoop 3.1.0. 
We need to either upgrade to Hadoop 3.2 or get the fix backported to Hadoop 3.1 
line.

 

> Migrate to listStatusIterator in moving files
> ---------------------------------------------
>
>                 Key: HIVE-24717
>                 URL: https://issues.apache.org/jira/browse/HIVE-24717
>             Project: Hive
>          Issue Type: Improvement
>            Reporter: Mustafa İman
>            Assignee: Mustafa İman
>            Priority: Major
>              Labels: pull-request-available
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> Hive.java has various calls to hdfs listStatus call when moving 
> files/directories around. These codepaths are used for insert overwrite 
> table/partition queries.
> listStatus It is blocking call whereas listStatusIterator is backed by a 
> RemoteIterator and fetches pages in the background. Hive should take 
> advantage of that since Hadoop has implemented listStatusIterator for S3 
> recently https://issues.apache.org/jira/browse/HADOOP-17074



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to